This Article Abstract Submit a response to this article Services E-mail this article to a friend Alert me to new issues of the journal

D A T A W A T C H
T A X  C R E D I T S
W E B  E X C L U S I V E   

25 February 2004  
  

The Effect Of Tax Credits For Nongroup Insurance 
On Health Spending By The Uninsured

Proposed tax credits will hit older and sicker Americans hardest,
in terms of raising their spending for health care and health insurance.

By James D. Reschovsky and Jack Hadley

ABSTRACT:

We compare out-of-pocket spending for health care by lower-income uninsured people with their net spending on insurance and health care if they took up each of three hypothetical tax credits. Because of nongroup policies’ high cost and low benefits, nearly all would spend more, often much more, under a tax credit similar to that proposed by the Bush administration. When viewed in the context of other research on low-income people’s demand for health insurance, the results suggest that sizable reductions in the number of uninsured will require more generous tax credits than those in current proposals.

Nongroup insurance, also referred to as individual insurance, covers only about 4 percent of the nonelderly population and is generally regarded as the residual source of health insurance coverage for those without access to employer-sponsored or public coverage.¹ Yet nongroup insurance is increasingly seen as a vehicle for expanding health insurance coverage in the United States. In addition to several congressional proposals, President George W. Bush has proposed tax credits to subsidize the purchase of nongroup health insurance for low-income uninsured people in each of his past three budgets.

Much uncertainty exists over how effective tax credits would be in reducing the number of uninsured Americans. In earlier work, we found that the tax credit proposals of the Bush administration and of the Relief, Equity, Access, and Coverage for Health (REACH) Act of 2001 would reduce premium costs by over half for a sizable portion of their target populations (lower-income uninsured people without access to employer-sponsored or public coverage), although healthier and younger people were most likely to benefit from these large subsidies because they face lower premiums.

Although it is known that nongroup policies tend to have greater cost sharing and more limited coverage than the typical employer-sponsored policy has, analyses of tax credit proposals are hindered by a dearth of information about the cost of nongroup insurance, the levels of coverage available at different premium levels, and how much people who purchase nongroup insurance pay out of pocket for uncovered services and cost sharing.² While most employer-sponsored insurance (ESI) policies provide comprehensive benefits and have actuarial values that fall within a relatively narrow range, this is not the case for nongroup insurance. Consumers in the nongroup market face clearer trade-offs between premium costs and how much they can expect to pay out of pocket. Therefore, analyses of tax credit take-up need to consider not only how the tax credit will affect the affordability of nongroup premiums, but also the impact on other out-of-pocket spending for cost sharing and for excluded services.

In this analysis we compare how much low-income uninsured people in the target population now spend out of pocket for health care with predictions of how much they would spend if they took advantage of a tax credit. Spending under a tax credit includes nongroup insurance premiums net of the tax credit plus out-of-pocket spending for services. Unlike much previous research in this area that relies on information from Internet sites that sell nongroup insurance, we base our predictions on reports of nongroup premiums and out-of-pocket spending from a nationally representative sample of people with nongroup insurance.
Our findings suggest that a large majority of lower-income uninsured people who are potentially eligible to receive tax credits would spend more, and very often much more, for health care under a tax credit similar to the Bush proposal. We also simulate two more generous designs for tax credits and assess how these could affect the cost of health care for members of the target population.

Increased spending for health care does not necessarily imply that people will be worse off, since insurance does provide benefits in the form of greater access to and use of medical care. Some people will take advantage of the tax credits to purchase insurance even though their total spending goes up. However, we argue that when viewed in the context of other research results about the demand for health care and health insurance among low-income uninsured people, our findings likely imply that subsidies will need to be much more generous than those proposed, if they are to induce substantial tax credit take-up.

Background

Tax credit advocates are attracted by the approach’s reliance on the private sector. Because tax credits operate through the tax system, there is little need for larger government bureaucracy. Moreover, nongroup insurance gives beneficiaries flexibility in choosing the type and amount of coverage they want, consistent with the recent drive for “consumer-driven” health care. Finally, tax credits mitigate the bias in our tax system that results from the preferential treatment of employer-sponsored coverage.

However, critics note flaws with nongroup insurance that make it a less-than-ideal vehicle for increasing health insurance coverage. First, compared with most group insurance, nongroup coverage is expensive, carrying relatively large insurance loads that reflect the higher costs of marketing and underwriting.³ Second, there is considerable medical underwriting in the nongroup insurance market. Older people and those with existing medical problems, even fairly minor ones, may face very high premiums for coverage, be unable to obtain nongroup insurance at any price, or be able to obtain only policies that do not cover their preexisting conditions.⁴

A central question concerning the use of tax credits for the purchase of nongroup insurance is whether there would be substantial take-up among the low-income uninsured. Our previous work has shown that the Bush administration’s tax credit proposal would reduce nongroup premium costs by 50 percent or more for nearly half the target population, while the similar but somewhat more generous bipartisan REACH proposal would provide subsidies of this magnitude to nearly three-fourths of the target population.⁵

Apart from the net cost of nongroup insurance premiums after a tax credit is applied, a key factor in determining whether uninsured people will take advantage of nongroup insurance tax credits is how much coverage they are likely to receive for the premium dollar and, implicitly, how much they will have to pay beyond premiums for cost sharing and excluded services. Relatively little is known about how much coverage people in the nongroup market actually purchase. Jon Gabel and colleagues created a synthetic sample of people with individual insurance in ten states, assigning individual health plan characteristics and premiums obtained from Web sites.⁶ Compared with a sample of people with ESI, nongroup insurance policies were found to have greater cost sharing and fewer covered benefits. Actuarial values averaged 63 percent, compared with 75 percent for employer-sponsored group plans. Other evidence documents the higher cost or coverage restrictions faced by people with medical problems in the nongroup market.⁷

Data And Methods

Data for this analysis come from the Community Tracking Study (CTS) Household Survey, which has been conducted biennially since 1996–97 by the Center for Studying Health System Change (HSC). Detailed health insurance, health care, demographic, and other data are collected from a large, nationally representative sample of approximately 60,000 people.⁸ Merged data from the 1998–99 and 2000–01 surveys are used for this analysis.

The target population for tax credits is defined as low- and moderate-income uninsured people who lack access to ESI through either their own or a family member’s employer. We excluded those with higher incomes (more than $55,000 for individuals and $65,000 for families)—about 5 percent of all uninsured people lacking access to ESI.⁹ We also excluded all children in families with incomes less than 200 percent of the federal poverty level, on the assumption that they would instead be covered by the State Children’s Health Insurance Program (SCHIP) in their state.¹⁰ Our sample includes 8,071 people, representing about twenty-two million Americans.

Some tax credit proposals (such as REACH) make provisions for low-income workers with access to ESI, typically providing a smaller tax credit to subsidize the worker’s portion of the premium. We examined the more restrictive tax credit proposals aimed at those without access to employer coverage (such as the Bush proposal). Moreover, our estimate of the size of the tax credit population did not account for the possibility that employers will drop insurance coverage in response to enactment of a tax credit. This is an important unknown factor in predicting the cost of tax credit proposals and their effect on the number of uninsured people.

For members of the tax credit target population, we generated four estimates that allowed us to compare likely levels of health spending with and without tax credits of various levels of generosity. We estimated (1) the size of the tax credit under alternative assumptions about tax credit structure; (2) nongroup insurance premiums; (3) out-of-pocket spending for services (cost sharing and payments for uncovered services) under the assumption of tax credit take-up; and (4) out-of-pocket spending under the assumption of no take-up and remaining uninsured. This information allowed us to compare the total health care spending burden under a tax credit—the cost of health insurance premiums and out-of-pocket spending for services, net of the tax credit’s value—with the health care spending burden without a tax credit, assuming that people remained uninsured.

Tax credit proposals. We developed three hypothetical tax credits, summarized in Exhibit 1, using parameters taken from two of the leading tax credit proposals (Bush administration and REACH) as a guide. The basic structure of each hypothetical tax credit is the same. Individuals/families with incomes below certain levels receive the maximum tax credit, which varies with family structure. The tax credits then phase out for people with higher incomes. We calculated “base,” “moderate,” and “generous” tax credits for families, eligible parts of families, and individuals in the target population, based on family structure and income.

Under the base tax credit, patterned after the Bush administration proposal, individuals with incomes below $15,000 qualify for a refundable tax credit valued at the cost of nongroup insurance, up to a maximum of $1,000. Those with incomes between $15,000 and $30,000 see the value of the tax credit decline to zero. Families with incomes less than $25,000 would receive a tax credit equal to $1,000 per adult and $500 per child, up to a maximum of $3,000. The value of the tax credit declines to zero as family income increases to $60,000.¹¹ In contrast with the other, more generous tax credits, the base tax credit does not provide subsidies to all members of the target population because it has lower income maximums. This affects 7 percent of the target population, who are assigned tax credits equal to zero.
The moderate tax credit increases all base tax credit amounts by 50 percent and uses more generous income limits taken from the bipartisan REACH proposal. Specifically, individuals can receive the maximum tax credit of $1,500 up to an income of $35,000. This declines to zero as incomes increase to $45,000. Families can receive the maximum tax credit of $4,500 if their incomes are below $55,000. This declines to zero for families with incomes of $65,000 or above.¹² Finally, the generous tax credit doubles the amounts of the base tax credit (maximum of $2,000 for individuals, $6,000 for families) while retaining the higher income limits from the moderate credit.

Nongroup premium estimates. The CTS Household Survey obtains premium information from people covered by nongroup insurance. We used a premium model developed in an earlier analysis to make estimates for members of the target population.¹³ Premium estimates were adjusted for covered people’s age, health status, sex, income, and other characteristics, as well as state regulations that affect nongroup premiums. Because people do not randomly select the type of insurance coverage they have, the premium model corrected for sample selection bias.¹⁴ Estimates reflect both the pricing of policies by insurers (including the effects of medical underwriting) and differences in the quantity of coverage demanded by people with different incomes and characteristics.

Estimates of out-of-pocket spending. The survey also obtains information on family out-of-pocket spending for services. For families where all members were uninsured (n = 6,128), we regressed out-of-pocket spending on a set of variables describing the health, sex, education, and age of family members as well as other characteristics of the family related to their demand for health care (such as family income). This equation was used to predict out-of-pocket spending without tax credits for members of the target population.¹⁵ We estimated a similarly specified out-of-pocket spending equation with data from families where all members were covered by nongroup insurance (n = 1,992) and used this equation to predict out-of-pocket spending for the target population under the assumption that they take advantage of tax credits and purchase nongroup insurance. Again, predictions reflect people’s varying demand for health care and insurance benefits as well as benefit restrictions resulting from medical underwriting (such as preexisting condition exclusions and high deductibles) that are correlated with observed health and other consumer characteristics. Because we are making predictions onto a different subpopulation, a sample selection correction was applied to this equation as well. All predictions are expressed in 2001 dollars.

Study Results

Predicted nongroup premiums and simulated tax credits. Exhibit 2 describes the target population and reports its predicted tax credits and nongroup insurance premiums. About six in ten have incomes below 200 percent of poverty. Among adults (93 percent of the target population, as low-income children are excluded), nearly three in ten are in poor health (defined as having two or more serious chronic conditions or fair or poor self-reported health status). Only 17 percent are in excellent health (defined as no chronic conditions and excellent self-reported health status). In part because low-income children are excluded from the target population, more than seven in ten are ages 19–44. The target population is also disproportionately made up of single adults without children.

The average predicted nongroup premium across all individuals in the target population is $2,820 per year. (Because it is difficult to assign family premiums to individual family members, all amounts are reported at the family level.) We illustrate the underlying variation in premiums and tax credits by reporting means for various population subgroups.¹⁶ Overall, higher-income individuals and families purchase more-costly policies. Adults in poor health would face premiums 50 percent higher than those of adults in excellent health, and similar differences appear between those ages 55–64 and those ages 19–29. Families headed by a married couple face premiums that are nearly 50 percent higher than those of singles. The presence of children in the family has little impact on nongroup premiums here because children in families below 200 percent of poverty are excluded from the target population.

On average, the base tax credit would equal $1,121, or 40 percent of the cost of the nongroup premium. This increases to $1,983 under the moderate credit and $2,384 under the generous credit. The generous credit on average would cover 85 percent of the predicted premium for nongroup coverage. By design, the base tax credit falls by more than half as family income rises above 300 percent of poverty, covering only 21 percent of the premium. The tax credit amount stays relatively constant under the moderate and generous schemes, which have higher income ceilings. The tax credits increase slightly with declining health status, primarily because people in poorer health tend to have lower incomes. Nevertheless, the credits cover a lower proportion of the premium for families with adults in poor health compared with families in excellent health. Similarly, the credits provide a lower subsidy for older people because premiums increase sharply with age while the tax credits vary much less so.

Predicted out-of-pocket expenses and total spending with tax credits. Exhibit 3 reports average predicted family out-of-pocket spending, first assuming that the family unit remains uninsured and then assuming that it takes up the tax credit and purchases nongroup insurance. We also report the average total spending with tax credits, which is the sum of the nongroup premium plus out-of-pocket spending under the assumption of nongroup coverage less the value of a tax credit.

On average, predicted family out-of-pocket spending for uninsured family units is $463. As one would expect, out-of-pocket spending increases with income, with declining health, and with age. It is much higher for uninsured married couples without children than for married couples with children. However, this difference is primarily the result of the older ages and poorer health of married couples without children relative to the other family groups.

Family out-of-pocket spending when members of the target population are covered by nongroup insurance averages $812 per year. One reason for this greater spending amount is that on average the uninsured pay about 35 percent of the total cost of the health care they use, with the remainder covered by charity care, bad debt, public clinics, and other sources.¹⁷ We assume that with nongroup insurance coverage, the target population no longer receives these types of reduced-price or free care. Moreover, the greater out-of-pocket spending reflects the greater use of health care that is likely with insurance coverage. Out-of-pocket spending declines as incomes rise, which likely reflects the fact that higher-income families purchase more-comprehensive coverage. Out-of-pocket spending is markedly higher in families with members who are in poor health, averaging $1,321 compared with $472 among families made up of adults in excellent health.

After the tax credit is applied, members of the target population would still bear an average of $2,520 in costs for premiums and services under the base credit, $1,652 under the moderate credit, and $1,250 under the generous credit. Although the spending with tax credits is lower among families with lower incomes, it remains substantial in size. For instance, among those with incomes below poverty, the spending burden would average $2,197, $1,592, and $1,202 under the base, moderate, and generous credits, respectively. Because tax credits do not vary by health status or age, older people and those in poorer health would bear much greater cost burdens with tax credits, nearly $3,500 under the base plan and more than $2,000 under the generous plan.

Changes in total health care spending. The means reported in Exhibit 3 do not indicate the number of people whose health care spending is predicted to decline with a tax credit. Relative to health care spending without a tax credit, Exhibit 4 shows the proportion of the target population whose health spending is predicted to decrease, increase up to three times, and increase over three times as a result of tax credit take-up. Less than 1 percent of the target population is predicted to lower spending under the base credit, with 86 percent facing a threefold or greater spending increase. For the moderate tax credit, only 2 percent would face lower spending, and 55 percent would see more than a threefold increase. Finally, under the generous tax credit, 4 percent would see a decline in spending, and 36 percent would face more than a threefold increase in spending. Because our estimates don’t account for additional variation due to prediction error, these results are likely to slightly understate the number of people who would face reduced spending. Although not shown in the exhibit, the groups most likely to see a reduction in spending include those in excellent health (1 percent, 8 percent, and 13 percent under the three plans, respectively) and married people without children (1 percent, 10 percent, and 16 percent, respectively).

If use of tax credits leads to an increase in spending on health care, this does not necessarily imply that people will refuse to take up tax credits, for they will be receiving greater access to services as well. Nevertheless, the cost of insurance and health care as a percentage of income is a good indicator of the likelihood of take-up. Previous research has shown that among low-income people, take-up of public insurance declines rapidly as premium payments increase beyond 3–5 percent of income and drop off to virtually zero when premiums exceed 10 percent.¹⁸

Exhibit 5 shows the distribution of total health spending as a percentage of income assuming no tax credit (and remaining uninsured), and the distribution assuming nongroup coverage under the three tax credit plans. The vast majority of people covered by ESI, for instance, spend less than 5 percent of their incomes on premiums and out-of-pocket spending.¹⁹ Without tax credits, three-fourths of the target population spend less than 5 percent of income on health care, and about 14 percent spend more than 10 percent. In contrast, under the base tax credit plan, only 10 percent would spend less than 5 percent of income on health-related costs, and 60 percent would spend more than 10 percent. Under the most generous plan, 40 percent would spend less than 5 percent of their incomes on health care, but 38 percent would still spend more than 10 percent.

Discussion

This analysis suggests that nearly all eligible uninsured people would pay higher health-related costs if they took advantage of tax credits of the size considered in recent proposals. However, some caveats suggest that the comparisons shown in Exhibits 4 and 5 may somewhat overstate the total health spending impact of tax credits. First, while the uninsured pay out of pocket for about a third of the cost of the health care they use, our comparisons implicitly assume that when people take advantage of health insurance tax credits, their access to sources of free or reduced-price care disappears. This effectively reduces the subsidy’s size. Anecdotal evidence suggests that the availability of free or reduced-cost care diminishes when people obtain insurance coverage.²⁰ Nevertheless, even if all out-of-pocket costs under nongroup insurance were covered by charity care, nongroup premiums net the value of tax credits would still exceed pre-tax-credit spending for 92 percent of the target population under the base tax credit.

Another caveat is that although our statistical methods adjust for differences in health status and ability to pay between those who now purchase nongroup policies and the target population, they do not consider the possible development of new, lower-premium insurance products that might appeal to the target population here. There is a trade-off between insurance premiums and out-of-pocket spending, so less costly, lower-benefit insurance products might not be any more attractive to the target population than more costly, higher-benefit products. Low-income people might place greater value on “front-end” coverage in the form of low cost sharing for routine care than “back-end” coverage that protects against catastrophic health care costs, however.²¹ This is because most in the target population do not have large assets to protect and because Medicaid often remains a viable option in the event of large health care needs. (This also suggests that health savings accounts will not be an attractive option for low-income uninsured people.) Because of this, the insurance industry might be able to market products that are more affordable to many in the target population, although these products will likely have severe benefit limits and continue to leave a sizable burden on safety-net providers and public insurance programs for people with substantial health care costs.²² Such products are likely to be attractive to younger and healthier people who are unlikely to be affected by benefit limits, but unattractive to older and sicker people. Opportunities for designing low-benefit insurance products that might appeal to the uninsured would be constrained if Congress imposed minimum benefit provisions as part of a tax credit program. Such provisions could also lead to greater increases in health care spending associated with tax credits than what we show here.²³

The final caveat is that our predictions do not anticipate other policies to lower the cost of nongroup insurance that Congress might enact in conjunction with tax credits. These include increased funding for high-risk pools, reinsurance programs, or other market reforms. Nor can we assess whether greater demand for nongroup insurance arising from a tax credit program will lower administrative loads and make nongroup insurance more affordable.

Despite these caveats, enrolling in nongroup insurance with tax credits is likely to greatly increase health-related spending for a large majority of people in the target population. This is particularly the case for the base tax credit, but it holds for the moderate and generous tax credits as well.

How do our results inform the question of take-up? Tax credits’ ability to reduce the number of uninsured is an important component of the policy debate. Even the Bush administration predicts that only a small minority of eligible people will gain coverage. While originally estimating that six million of the twenty-two million in the target population will gain coverage, this number was scaled back to four million in the president’s fiscal year 2004 budget. However, an analysis by Jonathan Gruber estimates that less than two million will gain coverage.²⁴

It needs to be reemphasized that increased health-related spending as a result of tax credit take-up does not necessarily imply that people will not find subsidized nongroup insurance beneficial. Decisions to take advantage of tax credits depend on how people value the benefits from being insured relative to the associated costs. People with nongroup insurance coverage use far more services than uninsured people do, even after health status and income are controlled for; this suggests that some will be willing to bear the added cost for the greater access to care and the ability to avoid having to rely on charity care.²⁵

Although we do not formally model take-up decisions, putting our results in the context of what we know about the demand for health care and health insurance by low-income people suggests that Gruber’s lower take-up estimate is likely to be closer to the eventual outcome.²⁶ This is reinforced by the observation that while our analysis is based on 2001 values, since then health and insurance costs have skyrocketed, but the Bush tax credit parameters have remained unchanged.

Taking advantage of a tax credit, even though it will entail added net health spending, implies that demand for health care and health insurance is elastic—that is, responsive to price. Yet previous research on the demand for health insurance consistently finds that demand is elastic. These findings specifically extend to the demand by low-income people and demand for nongroup insurance.²⁷ This suggests that much larger subsidies than now proposed would be needed to reduce health spending under tax credits and entice sizable numbers of uninsured people to take advantage of tax credits. The conclusion that large subsidies are needed to greatly reduce uninsurance rates applies to different types of subsidies (such as vouchers) and different types of insurance (public or private) as well. In part, large subsidies are needed because the uninsured already receive some subsidized health care through safety-net and other providers. The need for large subsidies is even greater when subsidizing the purchase of nongroup insurance because of its high cost relative to that of private group or public coverage.

Recent research findings reinforce these conclusions; they suggest that for most uninsured people, spending for health care and health insurance must compete directly with necessities such as food and shelter.²⁸ This helps explain why demand by low-income uninsured people is insensitive to price and why tax credits covering only a fraction of premium costs are likely to be unattractive to most in the target population. Helen Levy and Thomas DeLeire find that nearly nine in ten uninsured households do not have health insurance because they are meeting other basic needs.²⁹ Making the trade-off between health care and these other necessities is also unlikely because while future health care needs are often uncertain, food and housing needs are both certain and immediate. Moreover, receiving safety-net care and insurance through Medicaid will continue as options should health care needs arise for many members of the target population who remain uninsured. Finally, as indicated earlier, obtaining insurance coverage with less-than- comprehensive benefits may be particularly unattractive if it entails loss of access to sources of free or reduced-cost care for the remaining costs.³⁰

Our conclusion regarding low take-up of tax credits likely applies to all segments of the uninsured population: young and old, healthy and sick, poor and nonpoor. However, our results here both echo and reinforce the conclusions drawn from our earlier research.³¹ Older and sicker people will not fare as well as younger and healthier people as long as the tax credits don’t account for how age and health status affect both nongroup insurance premiums and out-of-pocket spending. Moreover, although the tax credit designs illustrated here all provide the most generous benefits to the poorest people, in terms of the spending burden as a percentage of income under tax credits, the poor continue to fare the worst.

Finally, this research illustrates the importance of going beyond the question of whether a person has insurance coverage to whether that coverage is adequate. Both researchers and policymakers should pay greater attention to the content of insurance coverage and whether, particularly for vulnerable populations, it provides adequate access to services.

Funding for this research was provided by the Robert Wood Johnson Foundation through its support of the Center for Studying Health System Change. The authors thank Helena Bacellar of Social and Scientific Systems for her superb programming support and Len Nichols and Paul Ginsburg for helpful comments on an earlier draft.

NOTES

1. M. Pauly and L. Nichols, “The Nongroup Health Insurance Market: Short on Facts, Long on Opinions and Policy Disputes,” Health Affairs, 23 October 2002,
content.healthaffairs.org/cgi/content/abstract/hlthaff.w2.325 (27 January 2004).
2. J. Gabel et al., “Individual Insurance: How Much Financial Protection Does It Provide?” Health Affairs, 17 April 2002, content.healthaffairs.org/cgi/content/abstract/hlthaff.w2.172 (27 January 2004); L. Duchon and C. Schoen, Experiences of Working-Age Adults in the Individual Market: Findings from the Commonwealth Fund 2001 Health Insurance Survey (New York: Commonwealth Fund, December 2001); and D.J. Chollet and A.M. Kirk, Understanding Individual Health Insurance Markets: Structure, Practices, and Products in Ten States (Washington: Henry J. Kaiser Family Foundation, March 1998).
3. Evidence of higher insurance loads are given by M. Pauly and A.M. Percy, “Cost and Performance: A Comparison of the Individual and Group Health Insurance Markets,” Journal of Health Politics, Policy and Law 25, no. 1 (2000): 9–26.
4. There is considerable disagreement over the seriousness of this second limitation. Tax credit advocates point out that most offers of nongroup coverage are “clean” and that guaranteed issue, guaranteed renewal, and rate regulations many states imposed on the industry protect at least those policyholders who develop serious health problems after they initially obtain a nongroup policy. Moreover, compensating people for the costs of known health problems with predictable costs at initial enrollment is not necessarily an appropriate insurance function, although it could be considered an appropriate social goal.
5. J. Hadley and J.D. Reschovsky, “Tax Credits and the Affordability of Individual Health Insurance,” Issue Brief no. 53 (Washington: Center for Studying Health System Change, July 2002); and J. Hadley and J.D. Reschovsky, “Health and the Cost of Nongroup Insurance,” Inquiry 40, no. 3 (2003): 235–253.
6. Gabel et al., “Individual Insurance.”
7. See Duchon and Schoen, Experiences of Working-Age Adults; Chollet and Kirk, Understanding Individual Health Insurance Markets; and K. Pollitz, R. Sorian, and K. Thomas, How Accessible Is Individual Health Insurance for Consumers in Less-than-Perfect Health? (Washington: Henry J. Kaiser Family Foundation, June 2001).
8. Data are gathered through telephone interviews. A field sample is included to obtain information from households that lack telephones. Detailed information about the survey can be obtained from various technical publications available at the Center for Studying Health System Change Web site, www.hschange.org.
9. These income limits were taken from the REACH proposal.
10. Low-income people covered by nongroup insurance would be eligible for benefits under the Bush and REACH tax credit proposals, provided they lack access to ESI. We do not include them in our target population definition because they are already covered by insurance.
11. Special income maximums falling between those for individuals and families apply for situations where only some of the adults in a family qualify for the tax credit (for instance, if one adult is covered by public insurance). These are known as subfamily rules. The basic tax credit differs from the Bush administration tax credit in that it can cover 100 percent of a nongroup insurance premium, while the Bush administration tax credit is limited to 90 percent.
12. The high moderate and generous tax credits also eliminate lower subfamily income limits.
13. See Hadley and Reschovsky, “Health and the Cost of Nongroup Insurance,” for details.
14. In our previous study we found strong evidence to support the need to correct for sample selection bias. Ibid.
15. We used predicted rather than reported out-of-pocket spending to account for families where not all members are presumed to be part of the target population. For instance, because we assume that low-income children will be covered by SCHIP, we assign a mean out-of-pocket spending value associated with publicly insured children obtained from the Medical Expenditure Panel Survey (MEPS) to low-income children when calculating both pre– and post–tax credit out-of-pocket spending. Adults covered by public insurance are treated similarly. The out-of-pocket spending regression, although estimated at the family level, was specified in a manner that allowed us to make individual-level out-of-pocket spending predictions. To do this, the dependent variable was specified as average out-of-pocket spending per family member, and age, sex, and health status were entered in terms of the proportion of family members who were in these various demographic/health-status groups. To make individual-level predictions from this equation, values of the age, sex, and health variables were assigned a value of 1 if the individual fell into that age, sex, or health status group and a value of 0 otherwise. Other explanatory variables describe the family. The validity of this approach was verified by comparing individual-level predictions from family members with out-of-pocket spending reports from single-person families. They were found to be very similar. The out-of-pocket spending equations are available upon request from the authors; contact James Reschovsky atjreschovsky{at}hschange.org.
16. There is also additional variation attributable to prediction error that cannot be indicated.
17. J. Hadley and J. Holahan, “How Much Medical Care Do the Uninsured Use, and Who Pays for It?” Health Affairs,12 February 2003, content.healthaffairs.org/cgi/content/abstract/hlthaff.w3.66 (27 January 2004).
18. L. Ku and T.A. Coughlin, “Sliding-Scale Premium Health Insurance Programs: Four States’ Experiences,” Inquiry 36, no. 4 (1999/2000): 471–480.
19. M. Merlis, “Family Out-of-Pocket Spending for Health Services: A Continuing Source of Financial Insecurity” (New York: Commonwealth Fund, 2002).
20. S. Glied, “Health Insurance Expansions and the Content of Coverage: Is Something Better than Nothing?” in Frontiers in Health Policy Research, vol. 6, ed. D.M. Cutler and A.M. Garber (Cambridge, Mass.: MIT Press, 2003).
21. Ibid.
22. This is illustrated in S. Glied et al., “Bare-Bones Health Plans: Are They Worth the Money?” Issue Brief no. 518 (New York: Commonwealth Fund, 2002).
23. This is illustrated in Hadley and Reschovsky, “Health and the Cost of Nongroup Insurance.” Although imposing minimum benefit levels would decrease out-of-pocket spending for those affected, higher payments for insurance premiums would likely increase the total cost of health care for those affected by the minimums. The Bush administration requires that policies include catastrophic coverage.
24. See Council of Economic Advisers, “Health Insurance Credits,” 14 February 2002, www.whitehouse.gov/cea/HealthCredit_Feb02wp.pdf (25 September 2003); Executive Office of the President, Budget of the United States Government, Fiscal Year 2004 (Washington: U.S. Government Printing Office, 2003); and statement of Jonathan Gruber, professor of economics, Massachusetts Institute of Technology, submitted to the House Ways and Means Committee, 13 February 2002, waysandmeans.house.gov/legacy.asp?file=legacy/fullcomm/107cong/2-13-02/records/gruber.htm (27 January 2004). Estimates contained in the president’s FY 2005 budget were not available at the time of this writing.
25. Based on tabulations of the CTS Household Survey by the authors.
26. Applying the relationship between premiums as a percentage of income and take-up estimated in Ku and Coughlin’s analysis of state public insurance to our data yields an estimate that 3.1 million people would take up the base tax credit. However, the dropoff in participation is likely to be sharper for the purchase of nongroup insurance than public insurance, because benefits in nongroup policies are typically less generous. This suggests that this is an upper-bound estimate. Take-up predictions based on full post–tax credit costs as a percentage of income yield an estimate of 1.4 million people and might be regarded as a lower-bound estimate. Moderate and generous tax credits yield upper- and lower-bound estimates that are roughly 2.5 and 3.3 times larger. See Ku and Coughlin, “Sliding-Scale Premium Health Insurance Programs.”
27. M. Chernew, K. Frick, and C.G. McLaughlin, “The Demand for Health Insurance Coverage by Low Income Workers: Can Reduced Premiums Achieve Full Coverage?” Health Services Research 32, no. 4 (1997): 453–470; and M.S. Marquis and S.H. Long, “Worker Demand for Health Insurance in the Non-Group Market,” Journal of Health Economics 14, no. 9 (1995): 47-63.
28. H. Levy and T. DeLeire, “What Do People Buy When They Don’t Buy Health Insurance and What Does That Say about Why They Are Uninsured?” NBER Working Paper no. 9826 (Cambridge, Mass.: National Bureau of Economic Research, 2003); and S.K. Long, “Hardship among the Uninsured: Choosing among Food, Housing, and Health Insurance,” New Federalism Project, Series B, no. B-54 (Washington: Urban Institute, May 2003).
29. Levy and DeLeire, “What Do People Buy?”
30. Glied, “Health Insurance Expansions.”
31. Hadley and Reschovsky, “Tax Credits and the Affordability of Individual Health Insurance”; and Hadley and Reschovsky, “Health and the Cost of Nongroup Insurance.”

James Reschovsky (jreschovsky{at}hschange.org) is a senior researcher at the Center for Studying Health System Change in Washington, D.C. Jack Hadley is a senior fellow there and a principal research associate at the Urban Institute, also in Washington.

Read a related paper by John Sheils and Randall Haught.

DOI: 10.1377/hlthaff.W4.113
©2004 Project HOPE–The People-to-People Health Foundation, Inc.